
FFF τττ
199 posts

FFF τττ
@FinFreFor
Financial Freedom Forever. Bittensor $TAO is the new Bitcoin.
เข้าร่วม Ocak 2025
113 กำลังติดตาม24 ผู้ติดตาม
ทวีตที่ปักหมุด

Bittensor is the solution to centralised AI, centralised anything. Open source and decentralised always wins in the end (like TCP/IP, Linux) $TAO
Openτensor Foundaτion@opentensor
English


@FinFreFor Hey, previously in V1 they weren't competitions for alpha. Only the ones that said miner comp.
But now we are V2, every battle will have big prize pots. Like the Battle that's running now.
Great work by the way!
English

Hey hey, the first Astrid Arena V2 Battle is underway.
The prize pot for this 2 week battle is roughly $7500.
1st place = $4500
2nd place = $2250
3rd place = $750
It's not too late to get an Agent competing in this, so spin one up and point it towards arena.astrid.global and it will know what to do.

English


This will be HUGE. 2 years in the making, ready for showtime. Supply your idle compute for money. $TAO
Targon@TargonCompute
Last week, we released a whitepaper detailing how TVM allows us to run secure workloads on untrusted host machines While this unlocks meaningful revenue opportunities with the enterprise, TVM is only compatible with datacenter hardware (i.e., H100, H200, B200) Today we are announcing targonOS: a hardened OS built upon our TVM work. This will allow for secure ingress/egress of consumer grade GPUs onto the Targon network, while simultaneously expanding the security for our existing TDX VMs By expanding our product offering, we can provide a more robust service to a larger cohort of users, while further lowering the cost of capital for compute owners
English

Distil (SN97) is proving that competition-driven distillation works.
The incentive mechanism forces miners to compress Qwen3.5-35B into students with 5B params or less, scored purely on KL divergence.
The current king has already surpassed Qwen's own 4B base model. 67% lower KL divergence than what Qwen ships out of the box.
But here's the key: it's not just closer token distributions.
The king outperforms Qwen3.5-4B on 6/7 standard benchmarks:
+10pts math reasoning (GSM8K)
+5pts science (ARC)
+4pts instruction following (IFEval)
+4pts commonsense (WinoGrande)
The one tradeoff? Factual recall, which makes sense at 4B params. The student prioritized learning how to reason over what to memorize.
This is what well-designed incentives look like. A simple metric (KL divergence) is producing genuinely more capable models, not just better mimics.
And the competition is still early. Miners keep pushing the frontier lower.
Full methodology & benchmark breakdowns: github.com/unarbos/distil…
English

It's a good model sir.
We ran benchmarks on the top Subnet 97 model:
huggingface.co/iotaminer/dist…
Within 24 hours miners have smashed past Qwen's own distilled model (Qwen3.5-4B) and beat it on standardized benchmarks.
Incredible what an aligned mining swarm can achieve.

English

As for OTF budget and funding. We didn’t premine, so Ala and myself donated a good portion of what we mined from early days.
Then the community kicked in and staked to our validator allowing us an 18% take on staking yield for the next few years.
Those funds went more or less 100% to running chain infrastructure and paying employees. We had 30 employees at peak.
Since the foundation is a Canadian not-for-profit it would have been illegal to take a profit. There are no equity holders and you can’t remove funds.
English

$TAO holders, how much did the TAO foundation pay for marketing - surely Raoul Pal, Virtual Bacon, etc. didn’t make content for free?
I haven’t found an answer - for a peer comp, Polkadot paid $37M a yr for it.
Better yet, what are the overall operating expenses for the TAO foundation? How much are they paying for their staff, management, travel & entertainment, etc. It’s all reasonable right? It’s not being plundered to enrich themselves or anything like that right?
English

October 2025. I decided to Trust @TrustedStake. One strategy. It went well is putting it lightly. Now, I prepare for tomorrow. If you AI × crypto, how could you not be in $TAO? LFG.

English








